DESCRIPTION (adapted from the Abstract): The primary long-term goal of this project is to build a computational knowledge base management system, (called 'NeuroScholar') that can be used by neuroscientists to build comprehensive, coherent, useful accounts of the published literature describing the neural basis of behavior at a systems level. These accounts will typically involve a large number of phenomena over a large number of brain structures. This task subdivides into five specific aims: (1) To develop a general framework for the knowledge-management of published information in general (involving automated communication with other databases). (2) To design and implement a formal computations approach to studying the neural basis of behavior by representing each of the various different types of neuroscientific papers that are relevant to this issue, including those that describe ethological, physiological, neuroanatomical and neurochemical experiments. (3) To make powerful statistical and mathematical methods available within the system that provide researchers with innovative methods to analyze and examine the information in the literature. (4) To implementation the system practically, so that it may be used via the Internet by trusted users. (5) To test NeuroScholar by constructing an account of the neural basis of an example behavior: defensive behavior in the rat. This subject has a sizable literature for which a definitive review, produced by the end of the proposed funding period with the NeuroScholar system, would elucidate the organization of neural circuits underlying aggression across the whole brain.